Our laboratory has isolated, cloned, sequenced, expressed and determined the chromosomal localization of mouse and human IA-2 and IA-2 beta, both of which have turned out to be major autoantigens in type 1 diabetes. The genomic structure of these molecules also has been determined and the 5' upstream regions sequenced and shown to possess promoter activity. Sequence analysis revealed that IA-2 and IA-2 beta are members of the protein tyrosine phosphatase (PTP) family, but lack enzyme activity because of the presence of two amino acid substitutions in the highly conserved catalytic domain. Correction of these substitutions by site-directed mutagenesis resulted in the restoration of enzyme activity. Detailed molecular and cellular biology studies have provided information on the processing and post-translational modification and antigenic determinants of IA-2. Our studies also have uncovered homologs of IA-2 and IA-2 beta in C. elegans and Drosophila. These and other findings argue that IA-2 and IA-2 beta belong to a new subgroup of the PTP super family. Perhaps the most important outcome from these studies is that autoantibodies to IA-2 and IA-2 beta appear years before the development of clinically apparent type 1 diabetes and, therefore, can serve as predictive markers for this disease. In collaboration with colleagues in the United States and England, we demonstrated this in three separate clinical studies. The first, with identical twins, showed that 90% of non-diabetic co-twins (of diabetic probands) that eventually went on to develop type 1 diabetes had autoantibodies to IA-2 five years or more years before the appearance of clinical disease. In contrast, none of the non-diabetic co-twins that failed to develop type 1 diabetes displayed autoantibodies to IA-2 at any time during the five-to-ten years of follow-up. The second study involved approximately 10,000 school children who were followed for up to 10 years for the presence of autoantibodies. Eleven of these children went on to develop diabetes and ten of them were found to have autoantibodies to IA-2 years before they developed type 1 diabetes. The third study involved close to 15,000 first degree relatives of type 1 patients who were followed prospectively. Of those who went on to develop type 1 diabetes, over half showed autoantibodies to IA-2 years before they developed clinical disease. Even more striking was the observation that in individuals who had autoantibodies to both IA-2 and glutamic acid decarboxylase (GAD), another major type 1 diabetes autoantigen, the likelihood of coming down with type 1 diabetes was approximately 50% at five years and was even higher at 10 years. In the general population it is estimated that nearly one in 400 individuals will at some time during their life develop type 1 diabetes. Thus, 40,000 children would have to be admitted to a clinical trial in order to study the efficacy of therapeutic intervention on the 100 children likely to develop type 1 diabetes. Now by screening populations for IA-2 and GAD, as few as 200 double-positive children (i.e., autoantibodies to both IA-2 and GAD) are all that are needed to obtain significant results, since 50% of these children are likely to come down with type 1 diabetes within 5 years. Similar observations now have been made in a number of other laboratories around the world and screening for double-positive individuals is being widely used to study the pathogenesis of type 1 diabetes and to determine who to admit into therapeutic intervention trials. Over the last couple of years we have focused on the function of IA-2 and IA-2 beta. We showed that targeted disruption of the mouse IA-2 gene resulted in alterations in glucose tolerance tests and impaired insulin secretion. We also succeeded in determining the complete genomic structure of mouse IA-2 beta which turns out to be very similar to the previously determined genomic structure of IA-2. We found that both genes consist of 23 exons. Exons 1-12 encode the extracellular domain, exon 13 the transmembrane region and exons 14-23 the intracellular domain. The introns of IA-2 beta vary in size from 751 bp (intron 18) to 146.7 kb (intron 2) and all the splice acceptor and donor sequences agree with the GT/AG rule. The major difference between the two genes is that IA-2 spans 20 kb, whereas IA-2 beta spans over 800 kb. The two gene thus differ in size by 40 fold. This is mainly due to five introns that are close to or over 100 kb in size. We also determined the genomic structure of human IA-2 beta and found that it was very similar to mouse IA-2 beta, but spanned close to 1000 kb. The underlying reason or mechanism for the enormous expansion of the IA-2 beta gene as compared to the IA-2 gene remains unknown, but may be related to unique discriminatory differences at the gene regulatory level. We now are in the process of knocking out the mouse IA-2 beta gene and hope to determine the IA-2 beta knockout phenotype.